What to do

with NS-2

Steffen Reidt

Outline

NS-2:
architecture,
features

Radio
propagation
models
What to do with NS-2
Mobility
models

My algorithms
Steffen Reidt
and protocols
evaluated with
NS-2
Our Cluster
Algorithm
Ph.D. Seminar Dec 2007
Other
application
that would
make sense /
no sense

Conclusion
 What to do
with NS-2

Steffen Reidt
Outline
Outline

NS-2:
architecture,
features

Radio
propagation
models

Mobility
• NS-2: architecture, features
models
• Radio propagation models
My algorithms
and protocols
evaluated with
• Mobility models
NS-2
Our Cluster • My algorithms and protocols evaluated with NS-2
Algorithm

Other • Other application that would make sense / no sense

application
that would
make sense /
no sense

Conclusion
 What to do
with NS-2

Steffen Reidt
NS-2: architecture
Outline

NS-2:
architecture,
Physical layer:
features
• Event/Packet based simulator:
Radio
propagation • smallest entity in NS-2 is a packet, which is modeled as an
models

Mobility
event, that is executed as certain point in time. Other
models events change the state of the nodes (movement, wake up,
My algorithms
and protocols
destroy)
evaluated with
NS-2 • all events are strictly ordered by time, ordered in a list.
Our Cluster
Algorithm • Sequential execution:
Other
application
• event is on top of event list
that would • (Change state of node or) decide if packet can be
make sense /
no sense transmitted from A to B:
Conclusion • Compute propagation power received by B from A.
• The transmission succeeds iff propagation power >
threshold value. (discrete function!)
 What to do
with NS-2

Steffen Reidt
NS-2: architecture
Outline

NS-2:
architecture,
features

Radio
propagation
models Other layers:
Mobility
models
• MAC: Basically 802.11, configuration possible (data rate,
My algorithms
preamble length), but no bandwidth grading
and protocols
evaluated with • Network: AODV, DSDV, OLSR, Flooding
NS-2
Our Cluster
Algorithm
• Transport: TCP, UDP, CBR
Other Much more protocol developed individually but not part of
application
that would NS-2 code.
make sense /
no sense

Conclusion
 What to do
with NS-2

Steffen Reidt
NS-2: architecture
Outline

NS-2:
architecture,
features

Radio • From the users perspective, NS2 is an OTcl interpreter

propagation
models that takes an OTcl script as input and produces a trace
Mobility file as output.
models

My algorithms
and protocols
evaluated with
NS-2
Our Cluster
Algorithm

Other
application
that would
make sense /
no sense

Conclusion
 What to do
with NS-2

Steffen Reidt
Outline
Outline

NS-2:
architecture,
features

Radio
propagation
models

Mobility
• NS-2: architecture, features
models
• Radio propagation models
My algorithms
and protocols
evaluated with
• Mobility models
NS-2
Our Cluster • My algorithms and protocols evaluated with NS-2
Algorithm

Other • Other application that would make sense / no sense

application
that would
make sense /
no sense

Conclusion
 What to do
with NS-2

Steffen Reidt
Radio propagation model
Outline

NS-2:
architecture,
features

Radio Radio propagation model:

propagation
models • Generally Free-space and Two-ray-ground model included
Mobility
models
in Open Source network simulators
My algorithms • Our implementation of an efficient ray optical radio
and protocols
evaluated with propagation model enables:
NS-2
Our Cluster
• Consideration of terrain and obstacles in simulations
Algorithm
• Incorporation of topographical data in routing protocols
Other
application
and other upper layer protocols
that would
make sense / • Interruption by buildings and reflection and deflection
no sense
effects are considered
Conclusion
 What to do
with NS-2

Steffen Reidt
Radio propagation model
Outline

NS-2:
architecture,
features

Radio
propagation
models

Mobility
models

My algorithms
and protocols
evaluated with
NS-2
Our Cluster
Algorithm

Other
application
that would
make sense /
no sense
Qualitative validation of the radio propagation model with
Conclusion
iNSpect [2]
 What to do
with NS-2

Steffen Reidt
Simulation of city area
Outline

NS-2:
architecture,
features

Radio
propagation
models

Mobility
models

My algorithms
and protocols
evaluated with
NS-2
Our Cluster
Algorithm

Other
Behavior of AODV under the consideration of buildings.
application
that would
blue = sending; yellow = forwarding; green = receiving
make sense /
no sense

Conclusion • But unrealistic mobility remains (nodes moving randomly

and ignoring obstacles)
 What to do
with NS-2

Steffen Reidt
Outline
Outline

NS-2:
architecture,
features

Radio
propagation
models

Mobility
• NS-2: architecture, features
models
• Radio propagation models
My algorithms
and protocols
evaluated with
• Mobility models
NS-2
Our Cluster • My algorithms/protocols evaluated with NS-2
Algorithm

Other • Other application that would make sense / no sense

application
that would
make sense /
no sense

Conclusion
 What to do
with NS-2

Steffen Reidt
Existing mobility models
Outline
Existing mobility models:
NS-2:
architecture, • Random mobility models (Random Waypoint, Random
features

Radio
Direction, Gauss-Markov): Do not provide realistic
propagation
models
mobility scenarios for tactical networks, where soldiers are
Mobility
moving in formations.
models
• Graph Model: Used to simulate especially vehicular
My algorithms
and protocols networks. Nodes are moving on predefined paths with
evaluated with
NS-2 autonomous speed and direction
Our Cluster
Algorithm • Obstacle Mobility Model: Nodes are tracing their
Other
application
predefined or randomly chosen goal on the shortest path
that would
make sense /
considering the obstacles
no sense
• Group mobility models: Rudimentary implementations
Conclusion
directly in NS-2 where a group is moving with constant
relative positions between the nodes ⇒ No possibility of
simulating tactical networks which are likely to change
their formation
 What to do
with NS-2

Steffen Reidt
Coalition Mobility Model
Outline

NS-2:
architecture,
features Coalition Mobility Model:
Radio
propagation • Combination of Graph and Group mobility model plus
models
formations modeling
Mobility
models • The logical center of the group traces one of the
My algorithms
and protocols
predefined paths
evaluated with
NS-2 • A group moves relative to the logical center and is likely to
Our Cluster
Algorithm change its formation
Other
application • Formations in our simulation for small unit light infantry
that would
make sense / are defined according to the FM 7-8 field manual [4]
no sense
• Realization as extension of the mobility model generator
Conclusion
CanuMobiSim [1] by functionality for group movements
 What to do
with NS-2

Steffen Reidt
Paths for groups
Outline

NS-2:
architecture,
features

Radio
propagation
models

Mobility
models

My algorithms
and protocols
evaluated with
NS-2
Our Cluster
Algorithm

Other
application
that would
make sense /
no sense

Conclusion

Defining paths for group mobility model

 What to do
with NS-2

Steffen Reidt
Organization of groups
Outline

NS-2:
architecture, • Hierarchical organization of groups enables definition of
features
elaborated mobility scenarios with low configuration effort
Radio
propagation
models
• Groups consist of nodes and subgroups which can change
Mobility
their formations (E.g a platoon consists of single nodes
models
such as the platoon leader and squads. The squad itself
My algorithms
and protocols consists of fire teams as subgroups and squad leaders as
evaluated with
NS-2
single nodes and the fire teams consist of 3 or 4 single
Our Cluster
Algorithm
nodes)
Other
application team leader team leader
that would
make sense /
no sense

Conclusion | {z } | {z }
wedge-left squad leader wedge-right

• Groups and formations can easily be defined in an XML file

 What to do
with NS-2

Steffen Reidt
Definition of a group in EBNF
Outline

NS-2:
architecture,
features

Radio
propagation
models

Mobility distance = ”real number”

models
angle = ”real number”
My algorithms
and protocols name = ”string”
evaluated with
NS-2 node = distance angle
Our Cluster
Algorithm formation = name {{node} {group}}
Other
application
group = name formation {formation} [distance angle]
that would
make sense /
no sense

Conclusion
 What to do
with NS-2

Steffen Reidt
Result: Improved simulation
Outline
scenarios for Tactical networks
NS-2:
architecture,
features

Radio
propagation
models

Mobility
models

My algorithms
and protocols
evaluated with
NS-2
Our Cluster
Algorithm

Other
application
that would
make sense /
no sense

Conclusion

Platoon crossing city area

 What to do
with NS-2

Steffen Reidt
Outline
Outline

NS-2:
architecture,
features

Radio
propagation
models

Mobility
• NS-2: architecture, features
models
• Radio propagation models
My algorithms
and protocols
evaluated with
• Mobility models
NS-2
Our Cluster • My algorithms and protocols evaluated with NS-2
Algorithm

Other • Other application that would make sense / no sense

application
that would
make sense /
no sense

Conclusion
 What to do
with NS-2

Steffen Reidt
Cluster algorithm:
Outline

NS-2:
• Probabilistic:
architecture,
features
• Zongpeng [5]: Probabilistic Power Management in
Radio
WANETs
propagation • Send Hellos in 0.2 sec to determine number of neighbors
models
• Own decision to be CH with prob ∼ 1/number of
Mobility
models neighbors
My algorithms • Deterministic:
and protocols
evaluated with • Amis [3]: Exchange information possibly in several rounds
NS-2
Our Cluster
• Send Hello messages in 0.2 sec to know neighbors and
Algorithm
their Ids
Other
application • Every node chooses node with highest Id as his clusterhead
that would
make sense /
(and thereby forces this node to be clusterhead)
no sense Examination:
Conclusion
• Existing cluster algorithm change 10 to 50% of the cluster
heads after every hello message, depending on mobility.
• Sending Hello message in 0.2 sec frequency kills nodes in
NS-2 with standard settings in one to 3 minutes.
 What to do
with NS-2

Steffen Reidt
Our cluster Algorithm
Outline

NS-2:
architecture,
features Modification of Amis’ cluster algorithm
Radio
propagation
• Retained structure:
models • d-hop cluster
Mobility • no gateway nodes
models
• no partition in fix clusters, usual nodes connect
My algorithms
and protocols dynamically to cluster heads
evaluated with
NS-2 • ⇒ avoidance of communication overhead and chain
Our Cluster
Algorithm
reactions of cluster head changes
Other • Modifications:
application
that would • Replace node Id as decision factor about CH by Quality
make sense /
no sense factor
Conclusion • Increase the broadcast frequency to ≥ 8 seconds
• Incorporate mechanism to decelerate changes of CHs
 What to do
with NS-2

Steffen Reidt
Parameters to consider
Outline

NS-2: • Quality factor might consider:

architecture,
features • Trust
Radio
• Battery level
propagation • Certainty of connection (Distance)
models
• Degree (number of neighbors)
Mobility
models • Routing (use nodes which are already involved in routing
My algorithms process to save energy)
and protocols
evaluated with
• Bandwidth (Do not choose node working already on full
NS-2 capacity)
Our Cluster
Algorithm • Topography (Do not rely on a route that is likely to break
Other down due to an obstacle)
application
that would • Attack probability: type (tank, soldier); communication
make sense /
no sense activity
Conclusion • One metric for every property
• Combine the metrics to one quality factor ∈ [0, 1]
• Breach of side condition ⇒ quality factor = 0
 What to do
with NS-2

Steffen Reidt
Combination of metrics
Outline

NS-2:
architecture,
features
• Create one final quality factor as a linear combination of
Radio
the part metrics s.t.
propagation • Every loading factor ∈ [0, 1]
models
• Sum of loading factors = 1
Mobility
models • Example considering node Id and battery level:
My algorithms
and protocols
• Loadings: (1,0) (1/3, 2/3) (2/3, 1/3) (0,1)
evaluated with
NS-2
Our Cluster
Algorithm

Other
application
that would
make sense /
no sense

Conclusion
 What to do
with NS-2

Steffen Reidt
Distribution of data
Outline

NS-2:
architecture, • Application: Distributed key generation algorithm,
features
non-interactive hierarchical key generation
Radio
propagation
models

Mobility
models

My algorithms
and protocols
evaluated with
NS-2
Our Cluster
Algorithm

Other
application
that would
make sense /
no sense

Conclusion

Static hierarchical structure of platoon in city area.

 What to do
with NS-2

Steffen Reidt
Outline
Outline

NS-2:
architecture,
features

Radio
propagation
models

Mobility
• NS-2: architecture, features
models
• Radio propagation models
My algorithms
and protocols
evaluated with
• Mobility models
NS-2
Our Cluster • My algorithms and protocols evaluated with NS-2
Algorithm

Other • Other application that would make sense / no sense

application
that would
make sense /
no sense

Conclusion
 What to do
with NS-2

Steffen Reidt
Other application (simulations)
Outline
that would make sense / no sense
NS-2:
architecture,
features Yes: (Lower layers)
Radio • Routing algorithms under the consideration of application
propagation
models parameters (movement, terrain)
Mobility • Algorithms that are only dependent on relative but not
models
absolute timings. (Time for receiving, sending message is
My algorithms
and protocols device dependent.)
evaluated with
NS-2 • Protocols, that spread some information/knowledge
Our Cluster
Algorithm through the network. (trust models, cluster algorithms)
Other
application
Critical: (Application layers)
that would
make sense /
• The higher you get in the OSI hierarchy, the more the
no sense
mistakes from the physical layers might be multiplicated.
Conclusion
• The results from (data-oriented) applications that are
build on multi-hop communication (routing) are always
questionable. (Such as my simulation on the dissemination
of key material.)
 What to do
with NS-2
CanuMobiSim.
Steffen Reidt http://canu.informatik.uni-stuttgart.de/mobisim/.
Outline iNSpect.
NS-2: http://www.igd.fhg.de/igd-a8/de/projects/
architecture,
features mobsec/inspect.
Radio
propagation A. D. Amis, Prakash R., D. Huynh, and T. Vuong.
models

Mobility
Max-Min D-Cluster Formation in Wireless Ad Hoc
models Networks.
My algorithms
and protocols
In Proceedings of the 21st INFOCOM, pages 32–41, 2000.
evaluated with
NS-2 Headquarters, United States Department of the Army.
Our Cluster
Algorithm Field Manual FM 7-8: Infantry Rifle Platoon and Squad.
Other
application
U.S. Government Printing Office, 1992.
that would
make sense /
no sense
Li Zongpeng and Li Baochun.
Conclusion
Probabilistic power management for wireless ad hoc
networks.
Mobile Networks and Applications, 10:771–782, October
2005.